[Printing-summit] Printer/driver testing and certification

[Robert L Krawitz]

   From: Ulrich Wehner <uwehner at lanier.com>
   Date: Fri, 18 Aug 2006 12:34:58 -0400

   how would you know that the printer drivers work?

   i guess on a simple one tray printer this is might be easy....

   However, on a printer with say 3-8 paper sources, some 30 supported
   paper sizes, several output options ranging from trays to booklet
   makers.....

   color, B/W, etc
   how about functions like locked print, secure print, user codes, etc?

   fax-drivers?

My suggestion would be to start by picking a standard set of use cases
of interest to a reasonably broad set of users.  For each capability,
evaluate a "support level" for each capability for each printer and
driver combination under test.

Support levels might be:

1	Printer capability fully supported by driver

2 	Printer capability partially supported by printer (describe
	exceptions/limitations)

3	Printer capability not supported by driver

4	Printer does not support capability (i. e. any use case that
	requires color output on a black and white printer would
	receive a 4 rating -- the printer is simply not appropriate
	for this use case).

Example use cases might include things like

* Print 10 page A4 document including text and color graphics at
  standard business quality

* Print 4x6 photograph on glossy paper

* Print label on CD

with appropriate specifications.

   is "some output was created" enough to verify?

We received a lot of complaints late in the Gutenprint 5.0 cycle about
CD's printing slightly "squashed" on a lot of printers, due to a
design decision we had made in the CUPS driver for how to handle full
bleed capable printers.  This would easily have passed the "some
output was created" test, but it wasn't meeting users' needs.

On a more basic level, we've had plenty of bugs involving something
that certainly created output, but the output was completely incorrect
for the printer in question -- the printer might or might not print
anything, or spit out paper, or it might work in color mode but not
black and white.  Realistically, somebody's going to have to evaluate
the output to decide whether the test passes or not.

Testing photo printers opens an even bigger can of worms.  People will
have certain expectations of color fidelity.

Based on the discussion in Atlanta, I think it will probably be
easiest to test business-oriented use cases than "artistic" use
cases.

-- 
Robert Krawitz                                     <rlk at alum.mit.edu>

Tall Clubs International  --  http://www.tall.org/ or 1-888-IM-TALL-2
Member of the League for Programming Freedom -- mail lpf at uunet.uu.net
Project lead for Gutenprint   --    http://gimp-print.sourceforge.net

"Linux doesn't dictate how I work, I dictate how Linux works."
--Eric Crampton